There are some studies that talk about the tension between open ended discovery and more carefully constructed teacher led progression of experiences.
Which might be through example projects shown, unplugged activities, cross curricular links - but this strays from the idea that children are self directed. As they are self directed within the ideas put forwards.
(e.g. start with a simple single sprite animation (to learn about sequence), then animations with multiple sprites (for concurrancy), then move shape activities (to say learn about repetition), then onto say quizzes (to learn about selection and simple variables), finally games (which often require combinations)).
Some studies suggest that pure discovery might lead to extremely fine grained programming or even ‘smelly code’ and question impact of minimally guided learning on underlying understanding of concepts and progression. BUT more research is needed to compare open ended to blended (such as code reading, guided exploration, debugging tasks, live coding, apprenticeship, including design, learning about the process of making etc)
Intructionist approaches may not be meaningful to the learner and may not be motivational. And highly directed teaching such as copy code - may lead to learned helplessness etc etc.
Clearly there is a tension here, which is difficult to navigate.
Meerbaum-Salant, O., Armoni, M., & Ben-Ari, M. (2013). Learning computer science concepts with scratch. Computer Science Education, 23(3), 239–264.
Meerbaum-Salant, O., Armoni, M., & Ben-Ari, M. (2011). Habits of programming in Scratch. Proceedings of the 16th annual joint conference on Innovation and technology in computer science education (pp. 168–172). ACM.
Garneli, V., Giannakos, M. N., & Chorianopoulos, K. (2015). Computing education in K-12 schools: A review of the literature. Global Engineering Education Conference (EDUCON), 2015 IEEE (pp. 543–551). IEEE.
Grover, Pea, & Cooper. (2015). Designing for deeper learning in a blended computer science course for middle school students. Computer Science Education, 25(2), 199–237.
Werner, L., Denner, J., & Campe, S. (2015). Children programming games: a strategy for measuring computational learning. ACM Transactions on Computing Education (TOCE), 14(4), 24.
Hansen, A., Hansen, E., Dwyer, H., Harlow, D., & Franklin, D. (2016). Differentiating for Diversity: Using Universal Design for Learning in Elementary Computer Science Education. Proceedings of the 47th ACM Technical Symposium on Computing Science Education (pp. 376–381). ACM.